A5010124145- Reinforcement Learning in Robotics
- Neural dynamics and brain function
- Optical Imaging and Spectroscopy Techniques
- Laser Design and Applications
- Advanced Memory and Neural Computing
- Computability, Logic, AI Algorithms
- Spectroscopy Techniques in Biomedical and Chemical Research
- Neural Networks and Applications
- Advanced Scientific Research Methods
- Cognitive Computing and Networks
- Slime Mold and Myxomycetes Research
- Economic and Technological Systems Analysis
- Data Mining Algorithms and Applications
- Infrared Thermography in Medicine
- Fluid Dynamics and Turbulent Flows
- Laser Material Processing Techniques
- Optical Coherence Tomography Applications
- Embodied and Extended Cognition
- Complex Systems and Time Series Analysis
- Robotics and Automated Systems
- EEG and Brain-Computer Interfaces
- Gene Regulatory Network Analysis
- Rough Sets and Fuzzy Logic
- Distributed and Parallel Computing Systems
- Cognitive Science and Mapping
Moscow Institute of Physics and Technology
2018-2024
P.N. Lebedev Physical Institute of the Russian Academy of Sciences
1992-2020
Institute of Physics and Technology
2018
Applied Mathematics (United States)
2018
Research Center of Neurology
2001
Time-dependent functions giving distributions over instantaneous values of electron momenta and energies in a laser field are derived. These distribution found for arbitrary the ratio thermal velocity to oscillation velocity. They necessary studies laser-radiation interaction with plasma, laser-plasma x-ray lasers, etc.
The goal here is to identify key directions for the future advanced research initiatives in Artificial Intelligence (AI) and beyond. following areas are identified as having particular importance: (1) socially emotional, ethical, moral AI, (2) self-developing self-sustainable (3) human-analogous inspired by human psychology. As a result, general concept formulated with intent clarify unify currently popular slogans, including General (AGI), Strong Human-Level or Humanlike AI (HLAI),...
A new approach to language modelling is proposed, Deep Structure Learning, orders of magnitudes faster than the existing neural networks based models due effective sparse (symbolic) coding. Language model consists a hierarchy computational layers, extracting structural relations between increasingly abstract sets symbols. The complexity learning scales linearly with size model, contrary quadratic scaling backpropagation learning.
Optical scattering spectra obtained in the clinical trials of breast cancer diagnostic system were analyzed for purpose to detect dataflow segments corresponding malignant tissues. Minimal invasive probe with optical fibers inside delivers white light from source and collects while being moved through tissue. The sampling rate is 100 Hz each record contains results measurements scattered intensity at 184 fixed wavelength points. Large amount information acquired procedure, fuzziness criteria...
This paper describes the results of using ADAM (Adaptive Deep Autonomous Machine) architecture in task predicting movements financial markets. In addition, data transform algorithm has been developed that converts their distribution over each coordinates similar to a uniform one.
We present Deep Control - new ML architecture of cortico-striatal brain circuits, which use whole cortical column as a structural element, instead singe neuron. Based on this architecture, we MARTI model human brain, considering neocortex and basal ganglia. This is de-signed to implement expedient behavior capable learn achieve goals in unknown environments. introduce novel surprise feeling mechanism, that significantly improves reinforcement learning process through inner rewards. OpenAI...